Why Your Table Saw Burns, Drifts and Kicks Back: Symptom to Cause to Measured Fix
Burn marks come from four things, in this order of likelihood: pitch on the blade, a fence toed in toward the blade at the back, a blade not parallel to the mitre slot, and too many teeth for the feed rate. Check them in that order, because the first costs nothing and fixes maybe a third of cases. Blade-to-slot wants 0.002 in or better over the blade diameter.
Four causes account for nearly every burnt edge. Pitch and resin baked onto the carbide, a rip fence toed inward at the outfeed end, a blade that is not parallel to the mitre slot, and too many teeth being fed too slowly for the cut.
Check them in that order. Cleaning the blade takes twenty minutes, costs the price of a tub of cleaner, and resolves a surprising share of burn complaints outright. There is no sense taking a dial indicator to the trunnions when the blade has a brown film on the shoulders.
If the blade is clean and the burn is still there, the number you are chasing is 0.002 in. That is the working tolerance for blade-to-mitre-slot parallelism measured on a single marked tooth at the front and back of the blade, and it is the check that everything else on the saw is measured against.
Clean the blade before you touch a single bolt
Resin builds on the carbide shoulders and just behind the teeth. It is not always obvious, particularly on a black-coated plate, so look at the blade under a raking light rather than from directly above.
A blade with pitch on it is running with an effectively wider, blunter tooth and a body that is dragging in the kerf. It burns cherry and maple immediately and it burns pine eventually.
Soak it in a purpose-made blade cleaner or a simple citrus degreaser, ten to fifteen minutes, then lift the softened deposits with a stiff nylon brush. Never a wire wheel: brass is borderline, steel will round the carbide edges and you will have converted a cleaning job into a resharpening job.
Dry it properly and put a light coat of dry lubricant or paste wax on the plate before it goes back on. A slick plate genuinely reduces friction burn on deep rips in dense hardwood.
Symptom, cause, and the number the fix has to hit
What you are seeing Most likely cause How to confirm it The fix, and the figure to hit Burn on both faces, worse toward the back of the cut Fence toed in toward the blade at the outfeed end Dial indicator riding the mitre slot, run along the fence face front to back Reset the fence parallel to the slot or 0.002 to 0.003 in open at the rear Burn on the offcut side only Blade not parallel to the slot; the back teeth are heeling into the cut Mark one tooth, measure it at the front and the rear of the slot Bring blade to slot within 0.002 in across the blade diameter Burn on straight rips of maple or cherry with a clean, sharp blade Too many teeth for the cut and too slow a feed Count the teeth. A 60T or 80T blade ripping 4/4 hardwood is the tell Fit a 24T flat-top or a 30T rip blade and feed steadily at 8 to 12 ft per minute Sticky brown film on the plate after two boards Pitch and resin, often from wet or resinous stock Look at the carbide shoulders under a raking light Soak and brush; check the stock with a moisture meter, above 12% burns everything The kerf closes behind the blade during a rip Internal stress or a moisture gradient releasing in the board Watch the kerf as the cut opens; it pinches on the plate Riving knife fitted and within 0.010 in of the blade plane; sticker the stock and re-cut Workpiece drifts away from the fence mid-rip Fence deflecting under hand pressure, or locked only at the front Push the fence face sideways at the rear with the lock engaged The fence must not move more than 0.005 in under firm hand pressure Crosscuts wander off square on the mitre gauge Side play between the bar and the slot Feeler gauges beside the bar at both ends of the slot Fit an adjustable bar or expansion screws; keep side play under 0.003 in Tearout on the underside of plywood and veneer Wrong tooth grind and no support at the cut line Look at the tooth face: flat-top grind on veneered sheet is the culprit ATB or Hi-ATB blade plus a zero-clearance insert sitting within 0.005 in of flush Fine ripples along the cut edge, evenly spaced Blade wobble from arbor runout or a bent plate Dial indicator on the plate, rotate the arbor by hand through a full turn Arbor flange under 0.001 in; blade plate under 0.005 in at the rim Motor bogs and the blade slows on a full-depth rip Belt slip, a dull blade, or voltage drop in the extension cord Check belt tension, then check the cord gauge and length 12 AWG cord under 50 ft. A 14 AWG 50 ft cord at 15 A drops about 3.8 V, over 3% of 120 V Cut edge square at the top, out of square at the bottom Blade not at 90 degrees to the table, or the insert sitting proud Engineer square against the plate between teeth, not against the teeth Reset the 90 degree stop to within 0.002 in over 3 in of blade height Kickback with no warning on a rip Riving knife removed, or fence toed in, or freehanding without the fence Look for the knife first; then measure the fence Refit the riving knife; fence parallel or 0.002 in open at the rear, never closed Find your symptom, do the confirmation step, and only then reach for a wrench. Changing two things at once is how a saw that had one fault ends up with three. The tolerance sheet
These are the figures a well-set-up saw holds. The good column is what a cabinet saw with proper trunnions should reach and hold; the acceptable column is where a contractor saw or a well-used machine can honestly live; the third column is where the fault is now causing the symptoms above.
Everything is measured with a dial indicator reading to 0.001 in, mounted on a bar or jig that rides the mitre slot. A magnetic base on the cast-iron top is not good enough for blade-to-slot work because it references the table, and the table is the thing whose relationship to the slot you are trying to establish.
What you are measuring How to measure it Good Acceptable Fix it now What it causes when it is out Blade to mitre slot, parallel Mark one tooth. Indicator on a slot-riding jig, read the same tooth at the front and rear of the blade 0.001 in or less 0.002 in Over 0.003 in Burn on one face, heel marks, kickback Fence to mitre slot Same jig, run the indicator along the full fence face 0.000 to 0.002 in open at the rear Up to 0.003 in open Any amount toed in Burn, binding, kickback Arbor flange runout Indicator tip on the machined flange face, rotate the arbor by hand 0.0005 in or less 0.001 in Over 0.002 in Blade wobble, ripples, bearing noise soon after Blade plate runout at the rim Indicator on the plate 1/4 in inboard of the gullets, full rotation by hand 0.003 in or less 0.005 in Over 0.010 in Oversized kerf, saw marks, burn on both faces Blade flatness, off the saw Straightedge across the plate on a known flat surface 0.002 in or less 0.004 in Over 0.006 in Wandering cut that no alignment will fix Riving knife offset from the blade plane Straightedge across the blade body and onto the knife Centred within 0.005 in Within 0.010 in Over 0.015 in Board catches on the knife, or gets no protection at all Riving knife thickness Calipers. Must be thinner than the kerf, thicker than the plate 0.005 to 0.010 in under the kerf Up to 0.015 in under Equal to or over the kerf Binding, or a false sense of safety Blade at 90 degrees to the table Engineer square on the plate between teeth, over 3 in of blade height 0.001 in or less 0.002 in Over 0.004 in Out-of-square edges, gappy glue-ups Throat insert flush with the table Straightedge across the table over the insert, both directions Within 0.002 in Within 0.005 in Over 0.010 in proud Workpiece catch, tearout, an unsafe stall Table and wing flatness 36 in straightedge and feeler gauges, checked in three directions 0.005 in or less 0.010 in Over 0.015 in over 36 in Rocking stock and inconsistent depth of cut Work down this sheet in order and stop at the first line you fail. Every check below a failed one is being measured against a moving target, so fixing them out of sequence wastes an afternoon. Order of operations, and why it is not negotiable
- Blade off. Clean it, check the plate for flatness and runout on the bench, and inspect the carbide for chipped or missing teeth. A bent blade makes every subsequent measurement meaningless.
- Arbor next. Runout on the flange with the blade off. If the flange is out, nothing you bolt to it will run true, and no alignment fixes it.
- Blade to mitre slot third. This is the master reference for the whole machine. On a cabinet saw you shift the trunnions; on most contractor saws you shift the table on its bolts.
- Fence to mitre slot fourth. The fence references the slot, not the blade, because the slot is the one thing on the saw that does not move.
- Riving knife or splitter fifth, aligned to the blade plane now that the blade plane is known and stable.
- Bevel stops at 90 and 45 degrees sixth. Both, and check the 90 stop again after you have set 45, because on many saws they interact.
- Insert and outfeed support last. Flush insert, outfeed table level with or a hair below the saw top, never above.
Write the readings on a card and date it. Six months later, when the saw starts burning again, the useful question is what changed rather than what is wrong, and you cannot answer that without the first set of numbers.
Taking a dial indicator reading that means something
The jig matters more than the indicator. A steel or hardwood bar milled to slide in the mitre slot without side play, with the indicator mounted rigidly to it, gives repeatable numbers. A magnetic base stuck to the table does not, and neither does a plastic bar with 0.010 in of slop.
Mark one tooth with a marker and use only that tooth. Blade plates vary in thickness across their diameter by more than the tolerance you are chasing, so measuring the front on one tooth and the back on another gives you the blade's manufacturing variation dressed up as an alignment error.
Preload the indicator about 0.020 in so the plunger is off its stop, always push the jig in the same direction along the slot to keep the backlash consistent, and take each reading three times. If the three do not agree within 0.0005 in, the jig has play and the numbers are fiction.
Rotate the blade backward to bring the marked tooth from the front position to the rear rather than moving to a different tooth. It sounds fussy. It is the difference between a real reading and a wasted afternoon.
Feed rate is a number, and it explains most burning
Chip load is what each tooth removes per pass, and it is feed rate divided by teeth per minute. Run the arithmetic once and the burn problem becomes obvious.
A cabinet saw with the arbor turning around 3,700 rpm and a 24 tooth rip blade presents 88,800 teeth per minute to the wood. Hand-feeding at a brisk 12 ft per minute is 144 in per minute, which works out at roughly 0.0016 in of chip per tooth.
Now put an 80 tooth crosscut blade on the same saw at the same feed rate. That is 296,000 teeth per minute and a chip load around 0.0005 in. Half a thousandth. At that point the tooth is not cutting so much as rubbing, and rubbing generates heat, and heat is the brown mark on your board.
This is why the fix for burning on rips is so often a blade change rather than an alignment. Fewer teeth, a flat-top or modified flat-top grind, a wider gullet to carry the chip, and a feed rate you maintain rather than hesitate through. Slowing down at the halfway point of a rip puts a burn mark exactly where you slowed.
Thin-kerf blades help on underpowered saws for the same reason: less material removed per inch of travel means the motor holds its speed and you can keep the feed up.
Kickback deserves its own paragraph
Kickback is the board being lifted by the rising back teeth and thrown back at you, and it happens for a small number of reasons that are all preventable.
The fence toed in at the rear traps the offcut between the fence and the blade. That is the single most common mechanical cause and it is why every fence spec above says parallel or slightly open, never closed.
A missing riving knife means the kerf can close on the plate. A riving knife rises and falls with the blade and stays the same distance behind it at any depth, which is why it is better than a fixed splitter, and there is no rip cut where removing it is a good idea.
Freehanding a rip without the fence, ripping stock with a bowed face against the table, or trapping an offcut between the blade and the fence on a crosscut are the operator causes. The mitre gauge and the rip fence do not get used together on a crosscut without a stop block that releases the workpiece before the cut starts.
Anti-kickback pawls are worth having and are not a substitute for any of the above.
When the saw is the problem and the setup is not
Some machines will not hold alignment, and that is a machine fault rather than a technique fault. The tells are consistent: a saw that reads 0.002 in on Monday and 0.006 in on Friday with nothing changed, a fence that reads parallel unloaded and moves 0.015 in when you lean on it, a stamped sheet-metal top that flexes when you set a straightedge across it, or trunnions that shift every time you raise and lower the blade through its full travel.
That is what the Trueness Index measures on a saw: how closely the machine holds its own published capacity, its arbor runout, its fence parallelism and its port airflow when the readings are taken in a working shop rather than off a spec sheet. A saw badged as a 3 hp cabinet machine that cannot hold blade-to-slot within 0.003 in through a week of use scores badly no matter what the label says.
If you are in that position, spend the money on the fence before the saw. An aftermarket fence on an average saw fixes more cuts than a new saw with an average fence, and it is a fraction of the cost.
If the trunnions themselves are the problem, though, there is no accessory that saves it, and every hour you spend realigning is an hour you are not getting back.